The Recognition Of Bird’s Nest On Transmission Line Towers Based On Video Image Processing Technology

Birds nesting behavior may affect the safe operation and easily result a short circuit or trip of the transmission line tower. In order to guarantee normal operation of electric transmission line, monitoring and early warning is a must.To solve this problem, this paper proposes a scheme based on digital image processing technology which can be used to recognize the bird’s nest behavior.This paper mainly studies the recognition scheme which includes three parts: image preprocessing, image feature extraction and recognition of classification machine. The scheme can not only deal with single picture and output the discriminant result, but also count the correct recognition rate of photo set. When it comes to the implementation procedure, the first step is to fulfill grayscale processing and interpolation conversion for collected birds’ nesting pictures. Then a connected region identification method for single tower images is used to indentify the existence of birds’ nests. For a large number of samples of the tower images, two kind of feature extraction methods (Garbor wavelet filter and PCA) are-used for birds’ nests indentification. In detail, the Gabor wavelet is used to extract the angle and scale information, then PCA is used for data compression. At last, BP, SVM and ELM are used as classifers for birds’ nests indeification. Particularly, the recognition accuracy influenced by different dimension of ELM and the sensitivity of the sample number on different algorithms are studied.The experimental results show that, in the machine recognition of large sample tower bird’s nest image, comparing to the BP neural network and SVM, ELM can obtain the higher recognition accuracy. Hence Gabor and PCA are choosed to extract the informaion of the bird nests images, and the ELM are used to indentify the existence of the bird nests in this paper.